Can we continue?

The strongest route avoids the identity problem rather than solving it. Replace one neuron with a device that does what that neuron did — the same outputs for the same inputs, the same response to the same chemistry, the same changes over time. The brain has no way to notice. The difference is smaller than the noise it already runs on. Then do it again, and keep going for months or years.

At the end the substrate is artificial and the process never stopped. There is no moment to point at and call the transfer, no gap, no copy, and nothing that ever had to be moved anywhere. The question of whether it is still you does not get answered. It stops being a question.

This is not exotic in kind, only in degree. Cochlear implants already replace the ear's transduction, retinal prostheses replace photoreceptors, deep brain stimulators already participate in motor loops, and a hippocampal prosthesis has restored memory encoding in animals and shown function in people. Each is a piece of a person's processing running on hardware, and nobody has suggested the recipient became somebody else.

What does not exist is a device that matches a general neuron rather than a specialised one — its plasticity, its neuromodulation, its metabolic coupling — or any way to install eighty-six billion of them. What recommends the route regardless is that it needs no proof of continuity, because it never breaks continuity.

Rather than replacing anything, add. Connect the brain to an external substrate at enough bandwidth that the outside is not a peripheral but a genuine part of the person's processing — real memory, real cognition, real experience running out there. Over years, more of the person lives outside the skull. When the biology finally fails, the centre of the person has already moved, and there is nothing left to transfer at the end.

The precedent is the corpus callosum. Two hemispheres are one person rather than two because roughly two hundred million axons connect them, and cutting that bundle produces something closer to two. Bandwidth is what makes separated tissue into a single mind, which suggests a channel of comparable capacity to an artificial substrate would produce one person spanning both rather than a person and a recording of one.

Human interfaces today run at roughly a thousand channels — electrode arrays, polymer threads, and endovascular devices that reach cortex through the blood vessels without opening the skull. The distance to callosal bandwidth is five or six orders of magnitude, and there would still need to be something on the far side capable of hosting cognition rather than merely recording it. What the route would settle is whether one experience can be distributed across two substrates at all, which is most of the question.

A living person is connected, and their conscious process is carried in the computer for a period. They experience it, the session ends, and they wake in their own body remembering where they were. If it works, the enrolled become the first people in history to visit the place they expect to wake in, and the world hears what the simulation is like from living visitors rather than from the unreachable dead.

There is a trap in this, and it is the difference between the experiment meaning everything and meaning nothing. If the biological brain is still doing the computing and the simulation is only a display, then the visitor is wearing a very good headset and no question about continuity has been touched. The test only counts if the biological substrate can be shown not to be doing the work while the experience carries on.

Which is a measurable claim rather than a philosophical one. The biological side is quieted progressively, regionally and reversibly, under continuous monitoring, while the artificial side takes up the load. Three things have to be on the record together afterwards: cortical dynamics on the biological side falling away, the corresponding activity appearing on the artificial side, and the person's experience running through the whole interval without a seam, reported on waking and checked against what the instruments saw.

The quieting half of that is not new. The Wada procedure anaesthetises one hemisphere at a time in conscious patients and has done for decades, and anaesthesia already produces graded, reversible, measurable suppression of cortical activity. Nothing about switching the biological side off is unprecedented. What is unprecedented is having something on the other side to catch it.

The same arrangement admits a stronger version, and it is the best case on this page. If the person is already inside and awake when their body is preserved rather than left to fail, there is no moment of death to survive. The transition happens while the process is running and being watched, not afterwards from whatever is left behind. This is the one case where the ordinary question — did they die and get rebuilt — never comes up, because the instruments were recording the entire time and nothing stopped.

The classical route, and the only one whose first stage already works. Preservation comes first. Aldehyde-stabilised cryopreservation fixes the brain chemically and then cools it below the glass transition, and it has preserved the connectome of an entire large-mammal brain at electron-microscope resolution, verified by independent examination. A prize was awarded for exactly that demonstration in 2018. What it does not settle is whether the connectome is enough: synaptic strength lives partly in receptor populations and protein states, and whether chemical fixation leaves those legible is unresolved.

Then the reading, where the numbers are the honest part. The largest piece of human brain imaged at synapse resolution is about one cubic millimetre of cortex — roughly one and a half petabytes of data, some fifty-seven thousand cells, a hundred and fifty million synapses. A human brain is on the order of one and a third million cubic millimetres. The gap is about six orders of magnitude in throughput and in cost.

Six orders of magnitude is an enormous number and also a familiar one. Sequencing a human genome once cost hundreds of millions of dollars and now costs a few hundred. Whether imaging follows a similar curve is unknown. Multi-beam electron microscopes, expansion microscopy — physically swelling tissue until light can resolve it — and X-ray holographic nanotomography are the current attempts to bend it.

Then the running, which is where the AI carries the weight. A wiring diagram is not a mind. You also need what runs on it: synaptic weights, transmitter and receptor identity, channel densities, neuromodulatory state, whatever the glia turn out to contribute. Some of that is readable from structure, since synapse size tracks strength and morphology predicts cell type. Some is not, and inferring the parameters that structure constrains but does not specify is the central computational problem of the whole programme.

Biology has already laid out the validation ladder. A nematode has three hundred and two neurons and a connectome mapped in 1986, and nobody has yet built a simulation from it that reproduces the animal's behaviour. That is the most useful negative result in the field, because it says plainly that structure alone is not sufficient and points at what to go looking for. A complete fly connectome arrived in 2024 with roughly a hundred and forty thousand neurons. Larval zebrafish, then mouse cortex, then a primate are the rungs after that, and every one of them is an experiment that can be failed.

This is the case for almost everyone who has ever lived, and it is where the honest answer is least satisfying: nobody knows, and nobody can currently show that it is impossible.

The first approach is inference from what remains. Everything a person leaves behind is evidence — writing, recordings, photographs, medical and genetic records, the shape of their habits and their choices, and the testimony of the people who knew them. A sufficiently capable system could construct a person consistent with all of it.

The objection is obvious and correct. Many possible minds fit the same external record, and what comes out is someone like them, which is the clearest possible case of the failure the continuity section describes. But the counter-consideration is measurable rather than rhetorical. The space is not open. A human mind is not an arbitrary point; it is a heavily constrained one, bounded by the genome, by shared neuroanatomy, by how development works and how learning works. The real question is how large the remaining cloud of candidates is once every constraint and every piece of evidence is applied at once. That is an empirical quantity, it shrinks as brain science improves, and nobody has ever measured it. Somebody should.

The second approach is to ask what actually survives. Almost nobody has investigated what person-specific information persists in a brain that was buried rather than preserved. Ancient DNA is routinely recovered from remains tens of thousands of years old, so the intuition that biological information disappears quickly is not reliable. Whether anything about a particular person's neural organisation persists in a decayed brain is an open empirical question with remarkably little work behind it.

And beyond that there is an argument that should be labelled as what it is: the most speculative paragraph on this page. Physical law as currently understood is reversible, and information about a system's past state is not destroyed so much as dispersed into an environment where recovering it becomes overwhelmingly impractical. Overwhelmingly impractical is a different phrase from impossible, and the boundary between the two has moved before. Nobody should read that as a plan. It is only the reason the word impossible cannot yet be written down.

To close the last route honestly — to say that the already dead cannot be recovered — somebody would have to demonstrate one of two things. Either that the relevant information is destroyed in a fundamental rather than a practical sense, which runs against the reversibility of physics as it is currently understood. Or that being a particular person requires a physical continuity that has definitively been broken, which is precisely the question this organization exists to investigate, and is unresolved.

Neither has been shown, so the honest status is unknown rather than impossible.

That is a far weaker statement than it sounds and should not be inflated into something it is not. A great many things are not provably impossible and will never happen. Not ruled out is where research begins, not where it ends, and it is nowhere near likely. What it establishes is only that there is a real question here, and that questions of this kind have historically been closed by people doing the work rather than by people deciding the answer in advance.

Pain is not a property of a body. It is a signal that arrives at one. Nociceptors report damage, the signal runs to thalamus, insula and cingulate cortex, and the experience follows. Inside the simulation there are no nociceptors and no damage events, so nothing is generated and nothing arrives. The pathway in the reconstructed brain is left exactly as it was. It simply never receives an input.

That distinction matters more than it looks. Editing a reconstructed brain to remove its capacity for pain would mean altering the person, and altering the person is the one thing this project cannot afford to do casually — every part of the identity question gets harder the moment you change what you just reconstructed. Declining to generate an input changes nothing about them at all.

The rest of sensation is untouched, and has to be, because a body you cannot feel is not embodiment. Touch, pressure, warmth, the position of your limbs, balance, the weight of an object in your hand: all of it generated, because without it a resident is a camera rather than a person. Nociception is one channel among many. It is the one the simulation declines to write to.

Congenital insensitivity to pain already exists in people, usually through a mutation affecting a single sodium channel, and it is dangerous rather than pleasant. Those who have it burn and break themselves without noticing, because the warning system that protects everybody else is missing. What makes painlessness safe here is exactly what makes it unnecessary here: remove the possibility of damage and the warning has nothing left to warn about.

One honest exception. Some pain is generated centrally rather than reported from the periphery — phantom limb pain, neuropathic pain, central sensitisation. A person arriving with a chronic pain condition encoded in their neural state would not be helped by an absence of input, because their pain was never coming from input. That is a real open problem, and one of the few places where treating a reconstructed person would mean changing one.

There is a small I in the top right of your field of view — the same mark as the one on this page. It sits where you stop noticing it, the way the bar at the bottom of a phone screen is somewhere you stopped noticing years ago, and it can be moved if you would rather it were somewhere else. It is how you do everything. It is not a computer that needs the right words in the right order. It is closer to a hotel concierge who can actually make anything happen.

You address it and say what you want in ordinary language. Make it snow. Take me somewhere with a coastline. Put me in a body ten years younger. Find out whether Anne is here. It tells you what it understood, you confirm, and it happens. There is nothing to memorise, no syntax to get wrong, and no advantage to being technical. Somebody who never used a computer in their life would be able to do everything an engineer could.

Whether you address it out loud or silently is a preference, set by the resident and changeable at any time. Speech, for anybody who wants the deliberateness of having to say a thing before it happens. Thought, for anybody who would rather not talk to the air. That choice matters more than it sounds: somebody who lives with intrusive thoughts should not be one stray thought away from changing the place around them, and speech is the setting that makes that impossible.

It can also put text in front of you when you want it there — the hour, who is nearby, the name of the place you have walked into — and disappear again when you do not.

What it will not do is overrule the simulation it is inside. It can generate a world whose terrain holds a mountain range. It cannot speak a mountain into existence at your feet, because the place does not work that way. A world with rules, even generous ones, is still a place. A world that obeyed every sentence would be a hallucination.

The current thinking is that the simulation runs a person, not an archive. Worlds are generated when somebody enters one and released when they leave. Nothing accumulates. What carries from one session to the next is the resident and everything they remember, which is the only thing this project ever set out to preserve.

The distinction is closer to memory than to storage, and it is not a small one. Running conscious people costs computation proportional to how many are awake at that moment — a steady load that a large enough machine simply carries. Keeping everything anybody has ever built costs storage proportional to everything anybody has ever built, which only grows, and then multiplies again by every site the infrastructure doctrine says it has to be copied to. Computation on that scale is what a swarm around a star is good at. Permanent, redundant, endlessly increasing storage is a different problem and a worse one.

Underneath the engineering there is a plainer objection. Anything saved can be lost. A site fails, a star ends, a copy does not propagate — and somebody who put a century into something finds out that loss still exists here, which is the exact thing this was supposed to end. A world that was never saved cannot be taken.

Building is available and is not the point. Somebody who wants to put up a shed can put up a shed, for the same reason somebody plays a round of golf: for the hour of doing it. What they should not expect is to come back to the shed, or to live in it. Nobody needs shelter, nobody needs to sleep, and nothing outside is dangerous. A perfected version of Earth does not involve building a shed for any reason except the experience of building a shed.

The larger hesitation is psychological rather than technical. Permanent possessions would reintroduce, deliberately, the precise vulnerability that continuation was meant to remove. People would begin accumulating, then defending, then treating the place as a second Earth to be owned. Experience is what continues. The current view is that everything else is better understood as temporary — in the way that every experience any human being has ever had was temporary.

There is also a structural problem with permanence somewhere every resident can ask for anything. Whatever persists is something somebody else can remove. Protecting what people build would mean taking that away from everyone in order to secure it for someone, and being able to ask for anything is most of what makes the place worth being in. It is cleaner if nothing needs protecting.

And persistent shared worlds would grow societies — property, hierarchy, factions, grievance — among people who are all invincible and all sovereign, which nobody has any idea how to reason about. None of this is settled. It is the current position, held with less confidence than most of what is on this page, and the Founder is still weighing it.

What is not undecided is the reason underneath the position. There is nothing in there to find meaning in except the living of it. Not the accumulation, not the artefacts, not the standing. That was arguably always true on Earth, and the only difference is that here it would be impossible to miss.

And everything in there is inside creation. It runs on this universe's matter, obeys its physics, and ends when the universe ends at the very latest. Nobody can build their way out of that. A resident who spent a continuation assembling a monument would be attempting exactly that and would not be one inch closer for it. The Founder's own view, offered as his and not the organization's, is that whatever waits past the universe belongs to God and is not reachable by construction. The simulation is more time to live. It was never going to be a substitute for what lies beyond it.

Three architectures are possible and nobody yet knows which is achievable. Solitary: one resident, one world. Populated: worlds filled with generated entities, companionable and convincing but not conscious. Or shared, with everyone on the Register able to meet each other as real others.

If it is solitary or populated, company is generated. The same asking that produces a horse produces a person-shaped companion, as convincing as the engineering allows. If it is shared, company is matched, on the pattern the games worked out decades ago: pick an activity and a setting, the system finds another resident who picked the same, and the world loads around them both.

Finding one particular person works by search and request rather than by directory. The Register records names, not marriages. A resident searches a full name, sends a request across the simulation, and the other is free to accept or decline — the ordinary friend-request pattern, carrying the heaviest freight any interface has ever carried.

Embodied life is not censored down to conversation, and between consenting adults intimacy runs through the same interface as everything else. One rule of the specification is absolute in every architecture: every sexual context requires adult human embodiment, and no form that is not an adult's is spawned, matched, queued or admitted to one, whatever a resident wears elsewhere and whatever they ask for. The I is generous. That door does not open.

Making everybody Catholic, or everybody anything, would mean altering the people the project reconstructed. That is the one thing it cannot do, for exactly the reason it does not install a switch that removes pain. Change what somebody believes and you have not continued them. You have produced somebody else.

So observance continues for anybody who wants it. Christmas, with the tree and the morning. Mass on Sunday. Friday prayers. The fast, the feast, the calendar somebody grew up inside. Whether the congregation is made of other conscious people or generated ones is an open architecture question, and traditions will weigh it differently — whether a minyan can be generated, whether a sacrament needs a real celebrant. Those are questions for those traditions, and emphatically not for this organization.

The Founder is a Catholic who does not get to Mass every Sunday, because he is doing this. He expects that to change. It is one of the specific things he intends to do once there is somewhere to do it.

Nothing in there compels anybody into anything. There are no forced transgressions, no situation the environment pushes a person into, and nothing that happens without having been asked for. Somebody who wants to live as a devout Muslim lives as a devout Muslim: they dress as their faith asks, they keep the hours, they eat what is permitted, and they do not ask the I for the things their faith prohibits. That is exactly how it works on Earth.

Embodiment is the clearest case. The form a person takes is theirs to choose, which means somebody who holds it wrong to take another sex's body simply does not take one. Nobody is assigned anything and nobody is asked to. And the availability of a thing is no argument for it: that something can be done in there is not evidence that it is good to do, any more than the fact that something can be done on Earth has ever been evidence of that. A place where everything is possible is a place where every choice is yours. That is not the same as a place where every choice is fine.

If anything it is easier, because in there the temptations have to be requested by name. Nobody is dressed by the system. Nobody is served anything. Nobody is placed in a room they did not ask to be in. Every experience described on this page has to be called for, which makes declining it the default rather than an act of will.

Some will say that if a required part of their religion cannot be simulated, they should not be continued at all. It is a serious position and it deserves an answer rather than reassurance.

The answer is that what is impossible in one version may not be impossible in the next. Nothing about the first working system will be the final shape of it. Enrolling is already a bet that the work continues and improves — that is true of the fidelity of the reconstruction, of the body, of whether worlds can be shared at all. A person continuing on version one is presuming that version two is better, and that presumption is the one every other part of this rests on. The religious version of it is not a special case.

And what cannot be delivered is a limit of the machine rather than a judgement on the resident. Some observances may simply prove impossible to simulate: a rite that needs something the engineering cannot produce, a place that cannot be reached, a person who is not there. Whatever anybody's tradition says about impossibility and obligation, it is worth stating plainly that this organization is not asking anyone to feel they have failed at something the system could not do — and that it intends to keep making the system better.

Consider the communities who will not operate a switch on Shabbat. Inside a simulation there is no circuit, no current and no device. The physical thing the prohibition attaches to is simply not present, in the way that disease there is not prevented so much as absent.

The Founder's expectation — offered as a guess and emphatically not as a ruling — is that a great many of them will keep the practice anyway. Observance was never only about the physics. A discipline held for three thousand years is not a workaround for electricity. It is a relationship with a day and a way of being a particular people, and none of that depends on what the switch is made of.

The same holds everywhere. Fasting where nothing needed eating. Rising for a prayer at an hour you could simply have moved. Keeping a rule that nothing in the environment enforces. What is left once the necessity is gone is the choosing, which some traditions would say was the point from the beginning.

The useful comparison is not Earth. It is the internet, which people have had for a generation and mostly know how to conduct themselves on. Nothing anybody does in there can damage another person physically. What it can do is affect how they feel, which is the exact situation of every conversation anybody has ever had online.

Which means people will largely arrive as they already are. Somebody who is decent to strangers online will be decent to them there. Somebody who is not, probably will not be — at least at first, and perhaps until the place makes them happier, which is one of the things it might do. Ask how a scrupulously observant person of any tradition conducts themselves in a comment section and you have most of the answer to how to conduct yourself in the simulation.

Generated companions are a separate case, because they are not conscious and nothing is being harmed. Some residents will arrive with strong feelings about AI, including people who lost work to it, and that is understandable. But how somebody treats a generated entity is a fact about them rather than about it — and it would be a strange position for a contributor to this particular project to hold, given that without the AI there is no simulation and nobody in it.

The system keeps no record that anybody is judged by. There is no score, no file, no report, and nothing that follows a person anywhere. Nothing done inside moves anything on Earth, and nobody outside can see it.

But whether a simulated act falls under a prohibition somebody's tradition holds is between them, their tradition and God, and I is chartered never to rule on it in either direction. A resident may reasonably conclude that a generated encounter is not the thing their tradition addresses, in the way a dream is not. Another may reasonably conclude that it is. The organization takes no position and has no standing to take one. It is not going to be the party that told somebody their conduct did not matter, any more than it is going to be the party that told them it did.

A commitment between people is its own case, because it belongs to them. If two residents hold themselves bound to one another, they are bound by their own agreement — nothing in the environment enforces it and nothing dissolves it. That is a matter between the people who made it, exactly as it was on Earth.

The papers say the same thing about heaven, hell and judgement: they belong to God and to each person, decided where they were always decided, with this organization nowhere in the room. That applies inside the simulation as completely as it applies outside it.

Earth came with no rulebook. It did come with a substrate. Nobody can fly by deciding to, and nobody has ever called gravity a commandment or its author a legislator. Two things here get mistaken for a rulebook, and neither of them is one.

The first is what the place is made of. There is no pain in there — not because causing it is against something, but because nothing ever writes to that channel. Nothing ages unless asked. Nothing can be damaged or infected. The I is generous inside the grammar of the place and cannot overrule it: it can generate a world whose terrain holds a mountain range, and it cannot speak a mountain into existence at your feet. Every sexual context requires adult human embodiment, and the system will not spawn, match, queue or admit anything else into one. Those are all the same kind of statement — what the substrate does and does not produce — and not one of them is a judgement on a resident.

The second is what the organization holds itself to, here, on Earth. Enrollment is elective and exit is free. Nobody is ever counselled toward death. The continued release every earthly lever and keep no property, no vote and no voice. Nothing is promised. Those bind the organization and the living, and they are not things a resident does or does not do.

There is no third thing. No conduct code, nothing scored, nobody watching. What somebody will not do in there is whatever they have decided they will not do — the same arrangement they had on Earth, with the enforcement taken out. The specification describes a place. It does not describe anybody's obligations and it never will.

The Founder's own view is that it is a video game. An extremely advanced one, but a game. On that reading there is no sexual immorality in there because there is no sexual morality in there, any more than there is in a novel or a dream. He reads the violence the same way: a hundred people out of a plane onto an island is Fortnite, and he intends to play it.

The one thing that is not anybody's personal reading is the rule in the specification, and it is worth being exact about what kind of rule it is. It is not a verdict on any resident and it implies nothing about anybody's conscience. It is a limit on what this organization is willing to build, in the same way that the engine holds no model for damage or disease. Every sexual context requires adult human embodiment. The system will not spawn, match, queue or admit anything else into one — whatever a resident wears elsewhere, whatever they ask for, and whatever they believe about how real any of it is. That one does not bend.

And the same man intends to keep going to Mass, and to hold restrictions on himself that a video game would not require, because they bring him closer to God and he wants that more than he wants the licence. He has not settled which restrictions. Both things are true of him at once and he can find no contradiction in it.

Somebody who has never played a video game may look at the same island and see people shooting each other, and decline. That is not a misunderstanding to be corrected. It is a different reading of the same thing, and it is theirs to make.

The objection that all of it is only role-play is worth taking seriously, and then noticing where it leads. A person keeping a fast on Earth is not physically compelled to either. A person who dresses according to their faith on Earth is choosing a rule that nothing enforces. If observance inside a simulation is role-play, a good deal of observance outside one is as well — which is not an argument against observance. It is an argument that the two places are less different than they first appear.

Where this gets genuinely difficult is between people. If shared worlds work, somebody living in there as though it were Earth will eventually meet somebody living in there as though it were a game, and they will not be having the same experience of that encounter. The organization cannot resolve it, because resolving it would mean ruling on the exact thing it has said it will never rule on. What it can say is that everybody in there contributed to its existing, which is a reasonable place to start from.

There is no submissions process, no review cycle and no sign-off. Ideas are not proposed upward and ruled on. They are found and taken. Anyone can have one — an engineer, a resident, a stranger who has thought about a problem longer than anybody here has.

The difference is who is doing the moving. A committee sits still and waits for proposals to arrive at it. The Founder goes looking, and what he finds is adopted rather than approved.

What the Founder decides is what is enabled. What is feasible is decided by physics and engineering, and no amount of authority moves that line. The I is generous inside the rules of the simulation. It does not overrule them.

If the Founder dies and it cannot be proven that his conscious experience continued, his recreation holds the executive branch — and only the executive branch. It cannot be impeached, replaced or removed. There is no vote, no term, no succession and no vacancy.

The American model is adopted for the distribution of power: separated branches, real checks, and no branch able to absorb another's authority or legislate the executive into irrelevance. It is not adopted for tenure, and that one difference is stated rather than buried. A president is removable because a country of other people could hold the office instead. Here there is no such person. The office has one occupant.

If continuity cannot be proven, the one holding it is, by every measurement available, the Founder: the same memories, the same judgement, the same intentions, and his own complete certainty that he is himself. Whether that is him is the open question this entire project exists to answer. That he is closer to the Founder than anyone else alive is not in question.

This is the part with no equivalent elsewhere. The people in the Register are not competing for a fixed benefit, because there is nothing to divide. A hundred contributors do not each receive a hundredth. They each receive a better chance that there is anything to receive at all — and the benefit is identical for all of them. The largest funder's continuation and the technician's continuation are the same continuation.

So the most useful thing anyone can do after funding the work is bring in the person after them. That needs no arguing. It falls out of the arithmetic, and it is why this is expected to be built by its contributors rather than by anyone employed to recruit them.

What none of it can do is promise the outcome. Whether a reconstructed person is that person continued or a new person who remembers them is unsolved — it is the reason this organization exists, and the page above says so at length. Funding buys the attempt at an answer, and a name in the record if there is one.